Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add antoinedc/MantaUI --skill manta-live-sessiongit clone --depth 1 https://github.com/antoinedc/MantaUIWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/antoinedc/mantaui/manta-live-session)<a href="https://agentmods.dev/skills/antoinedc/mantaui/manta-live-session"><img src="https://agentmods.dev/badge/skills/antoinedc/mantaui/manta-live-session/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/antoinedc/mantaui/manta-live-session"><img src="https://agentmods.dev/badge/skills/antoinedc/mantaui/manta-live-session.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00061 | $0.02272 |
| Opus 5 | $0.00030 | $0.01136 |
| Sonnet 5 | $0.00012 | $0.00454 |
| Haiku 4.5 | $0.00006 | $0.00227 |
Grade B, and why
manta-live-session scanned grade B with 2 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 11d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Asks for rootmediumPrivilege escalation
A mod that escalates privileges can change anything on the machine, not only the project.
It is **not** installed to `/usr/local/bin` because that needs sudo this box Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
their own `curl` calls against the box and quietly got them wrong — the RPC body How it starts
The opening of the file, as written. The whole thing — 190 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Live box access — manta-probe
What this is for
Some work cannot be verified by a unit test: a screen that only renders with a
real transcript, a card that only appears when a real turn blocks, a Simulator
that has to be paired to a real box. Before this skill existed, agents wrote
their own curl calls against the box and quietly got them wrong — the RPC body
shape, the auth header, the fact that the prompt endpoint is asynchronous,
and the fact that the pairing endpoint is loopback-only are each easy to miss,
and a half-working request looks exactly like a broken feature.
manta-probe is the one supported way in. Do not hand-roll curl against
the box; if this tool cannot do what you need, say so on the issue rather than
improvising.
The one rule you cannot break
Every prompt runs on voska/default. There is no flag, no argument and no
environment variable to change it — the model is a constant inside the tool. The
other providers on this box are metered and are not yours to spend. If you find
yourself wanting a different model, you are doing something this skill is not
for.
Where it runs
The box is the Linux machine alphaclaw — the same machine the Linux agents
already run on.
| You are | Command form |
|---|---|
a Linux agent (manta-dev, and anything on the box) |
~/.local/bin/manta-probe <cmd> |
| the macos agent | ssh dev@alphaclaw '~/.local/bin/manta-probe <cmd>' |
Always use the absolute path ~/.local/bin/manta-probe, never the bare name.
A non-interactive SSH shell has a minimal PATH that does not include
~/.local/bin, so a bare manta-probe fails with "command not found" over SSH.
This is verified, not theoretical — use the absolute form everywhere and the same
line works from both machines.
Commands
# The one-shot most tasks want: fresh session, prompt, wait, print the reply.
~/.local/bin/manta-probe ask "Reply with exactly one word: PONG"
# Keep the session afterwards (to inspect it in the app, or prompt it again).
~/.local/bin/manta-probe ask "Read AGENTS.md and summarise section 6" --keep
# Long-form: drive a session across several turns.
SID=$(~/.local/bin/manta-probe new my-label)
~/.local/bin/manta-probe prompt "$SID" "first turn"
~/.local/bin/manta-probe prompt "$SID" "second turn"
~/.local/bin/manta-probe messages "$SID" # dump the transcript
# Pair an iOS Simulator in one step — the app auto-claims this link, no typing.
~/.local/bin/manta-probe pair-url # manta://pair?box=…&code=…
~/.local/bin/manta-probe pair-url --simctl # the full xcrun simctl command
~/.local/bin/manta-probe pair # raw code + URL (physical phone)
~/.local/bin/manta-probe server # just the server URL
# Housekeeping — always finish with this.
~/.local/bin/manta-probe list
~/.local/bin/manta-probe clean
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 11d ago First seen · 190 lines · 61 tokens per session scan B 8046bf91081c
manta-live-session is a skill published in the GitHub repository antoinedc/MantaUI (12 stars, last pushed 4d ago), licensed MIT. It adds 61 tokens to every session and 2,272 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it B with 2 findings (asks for root, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
eide
A build tool for EIDE projects, an embedded-development extension for Visual Studio Code. It finds EIDE project settings and builds firmware using ARM CC or GCC.
workflow
A workflow skill for embedded-device projects that plans and runs building, flashing, debugging, observing, and diagnosis tasks.
im-adapter
Format responses for instant messaging platforms such as Lark, DingTalk, WeCom, Slack, and Telegram. Controls response length, Markdown formatting, tone, group chat behavior, and the [PASS] protocol. Also covers sending images and files via [SENDIMAGE]/[SENDFILE] markers on channels that support media (currently WeCom…
general
Handle everyday conversation, answer questions, manage files, take notes, run scripts, and maintain persistent memory across sessions. Use when the user asks a general question, requests file operations, wants to brainstorm ideas, needs to-do tracking, asks you to remember something, or requests skill search and…
message-push
Send proactive messages to IM groups or individual users via the gateway Send API. Use when the user says to send, post, notify, forward, or tell someone a message on Feishu, Slack, Telegram, Discord, DingTalk, or WeCom.
multi-bot
Coordinates responses between multiple GolemBot instances in a shared fleet. Use when the bot operates in a group chat with other bots, needs to decide whether to respond or pass, or must call a peer bot's API to fetch cross-domain data.